Posts: 51 From: Simsbury, CT 06070 Since: Feb 2001
posted 25 January 2002 10:26 AM
A thought about Andrew’s speculation regarding when PC’s will be ready for primetime videoconferencing, i.e. software codecs that can provide business quality videoconferencing. Andrew mentioned that today’s PC’s are "rated" at 1.8 Ghz. That is true, but depending on your configuration, the "real" output of some of these PC’s can be less than half of the "rated" performance. Here’s why. On a Dell 1 Ghz laptop with XP installed for example, the bios is by default configured so the machine runs, even with AC power, at 733MHz. Add to that the excessive background processing which occurs in XP and you may have a real performance rate between 450MHz and600MHz. If for example your PC runs on a corporate LAN running Windows 2000 or XP and you do not have "Administrator Rights", many actions on your PC are being constantly checked by both the internal PC and the server log files to see if you are allowed to perform a particular actions. This puts an extreme drag on performance. Again, the "rated" performance could be cut by about 30-40% in this case.
This is why a dedicated processor is still king when it comes to video processing. Will the 3-4 Ghz machines compensate? Possibly.
posted 26 January 2002 10:40 AM
Interesting analysis joe.... so the key question is when will software stop getting slower faster than hardware gets faster
posted 26 January 2002 11:30 AM
I think the decode performance is better than encode. Cuz we'll see that the Windows can play CD/DVD as good as we can imagine. Till now, it's good for stream video, just one way. For VC, we have to wait for a while...
posted 28 January 2002 12:37 PM
More thoughts from a WRB reader:
quote: Andrew: I enjoyed your take on software codecs as a disruptive technology. You stated that because "high performance machines will be available ... [in] ...2002" and thus "An ordinary PC with high-performance camera and display may be all that's required to meet customer expectations." I agree. You then stated that this was "A classic example of Christensen's market disruption." Here I think you get a bit ahead of the theory. Christensen's thesis is that disruptive technologies find a market for lower performance technology, and, in fact, don't meet customer expectations for the mainstream market. It's only after success with the other markets that the newcomers eventually overcome the old technologies.
So the question becomes, have software codecs already enjoyed success in other markets? Well, decoders have, as MPEG and MP3 browser plug-in engines can testify. Interactive codecs? Well, there's the booming success of the video chat room business... er, well, how the video mail we all keep getting...
I suspect that before software codecs start hurting the current market leaders, there will be a new market opened which can deal with low quality video. Perhaps video cell phones will establish a new least-common-denominator of quality, so the current SW-only desktop becomes good enough. Most likely, however, as you stated later in your article, the breakthrough will come when we realize the market is not for videoconferencing, but for other communication tools that happen to have video as a minor feature. That market will not be constrained by the traditionally slow videoconferencing growth rate.
posted 06 February 2002 06:59 PM
To become Andy's dream? true, I think there are two barriers.
1)OS (NT,XP ...) task switching & heavy GUI problem I think this is a problem Jodonnell mentioned here.
Solution) If VC makers uses non windows OS or use windows OS as booter, maybe near future PC has enough power to process encoding.It means use PC boards as a component of a dedicated system. There are methodology to run rtos on win-nt. I don't know it works for VC application or not.
2)VC to meeting solution. On business market, VC will become one of components of meeting solution systems. This is my forcast and some of other company.
In that case, main CPU should be busy to control power point or other application. In this case, it will never happened by software encoders, it always require one more CPU or DSPs.
Solution) But if someone use Tyan's motherboards, dual CPU, maybe software endoders may work well....
posted 11 February 2002 09:41 AM
I see the problems in the desktop market a little differently. We sell our wavelet-codec software, but not the 1.8 gigas are the problem but the connections. Quality in your company or university LAN is not the problem, you can get max. 640x480 pixels with our software but 75% of our customers ask to include modem or ISDN-users in their conferece. And this is the point where it gets ticky. So either you need a better compression or a good connection. So imo the real problem is not the processor (although it will solve the problem) but the connection. The processor is just the medium to handle the bandwidth problems.
Dietmar Fischer
Head of Sales and Marketing daViC GmbH Hoenower Strasse 35 / PF 16 10318 Berlin Germany Tel. +49/30/5019-2707 Fax. +49/30/5019-2566 Mail fischer@davico-gmbh.de
WR Forums
UBBFriend: Email This Page to Someone! 
posted 25 January 2002 10:26 AM
